How many pixels in a digital camera should be used to analyse one particle or character

How many pixels in a digital camera should be used to analyse one particle or character

Answer 1

[Jan]

What exactly is a "pixel in a digital camera"? You have to consider, that most digital cameras use different subpixels for different wavelength. Therefore the number of mega-pixel written on the camera is a artificial number only. And the actual resolution can differ from this mea-pixel-number remarkably.

What is a "particle" and a "character"? And what exactly means "analyse"?

[Jim]

what do you mean by different subpixels for different wavelengths?

particle is nothing but object.

How many minimum pixels in a digital camera should be used to identify one object?

[Walter Roberson]

http://en.wikipedia.org/wiki/Image_sensor#Color_separation

[Jan]

@Vivek: This depends on the object. It would be very helpful, if you explain any details. I have a camera with 0.001k-pixels to measure the brightness of a distinct area in front of my old manually controlled chemo-mechanic-camera. For this task 1 pixel is sufficient.

At work I have to identify reflective markers with a diameter of 9mm in a distance of 10m with an location error of 1mm. Then there cannot be enough efficient megas of pixels.

If you want to identify an ice bear in a snow storm, the number of pixels is less important than the accuracy for differentiating colors.

"Identify an object" is too general to be useful for a matching answer.

Answer 2

[Walter Roberson]

When analyzing an object (or character), you will have a minimum number of pixels needed to distinguish a feature of interest from other kinds of features. The number depends upon what you are looking for. The number of pixels needed to analyze a character in the "Latin Alphabet" (such as is used by English) is less than the number of pixels needed to analyze a Chinese character.

In my workplace, we would prefer to use images that exceeded 2048 pixels by 2048 pixels to analyze objects, but we find that in practice our current analysis algorithms cannot handle more than about 350 pixels per side in "real (enough) time".